Intro to 2D NMR: Homonuclear correlation COSY, lr-cosy, and DQ-COSY experiments

Transcription

1 Homework 9 Chem 636, Spring 2014 due at the beginning of lab April 8-10 updated 8 Apr 2014 (cgf) Intro to 2D NMR: Homonuclear correlation COS, lr-cos, and DQ-COS experiments Use Artemis (Av-400) or Callisto (Av-500) for this week s HW. Many choices exist for samples to be used for this and the following 2D labs: 1. A research sample is always OK (it should be a clean compound of moderate complexity). 2. our unknown compound: Many of these compounds are quite complex. High quality data will be obtained using the HW as guides. But these compounds are not good examples for first time attempts at using 2D spectra to assist with assignments. 3. Quinidine in C 6 D 6. This compound is of good complexity, and assignments are similar (but not identical) to those for quinine, as posted on-line. 4. Quinine in C 6 D 6. An on-line guide provides all assignments: Quinine and quinidine in C 6 D 6 are available in sample positions 2 and 3 on Callisto. Quinine and quinidine data are available on-line at: castor:\fry\public_html\chem636\quinine-quinidine 5. Sucrose or other samples used in previous homeworks. Reading Intro to 2D NMR: Claridge sections (esp. 5.1) COS flavors: Claridge sections 5.6, 5.7 Goals Learn basic terminology of 2D NMR, and the utility of various major flavors of COS (including TOCS) 2D spectra are acquired in a manner that is analogous to kinetic data sets that have many spectra that are often presented as stacked plots. ou will learn more about the second dimension in lecture. It is built experimentally by acquiring many one dimensional spectra with a delay in the pulse sequence, usually defined as t 1, being changed incrementally. The number of spectra that are acquired for the 2 nd dimension is set by the parameter TD1 (Bruker) [or ni (Varian)]. A COS pulse sequence has a 2 nd 90 pulse added to the standard one pulse 1D sequence: Standard COS sequence The delay between the two 90 pulses is t 1 ; on the spectrometer it is the delay d0 [or d2 (Varian)]. This delay will be increased incrementally as one 1D spectrum after another is acquired. d0=0 for

2 HW#9: 2D NMR, COS Pg. 2 the first spectrum, 2 nd spectrum has d0=1/sw1, 3 rd spectrum d0=2/sw1, and the n th spectrum d0=(n-1)/sw1. t 1 is 1 st time period, and is also called the indirect dimension; it is also called the evolution time. The 2 nd time period is t 2, and is the same normal acquisition period as used in 1D spectroscopy; remember that 2D is the acquisition of a series of 1D spectra. After two Fourier transform, one along each dimension, t 1 and t 2 are changed to F1 and F2, respectively. Resolution takes on different requirements in 2D spectra, usually in needing to only resolve different spins (or multiplets) from each other. The need to resolve within a multiplet is removed or reduced. Lower resolution is therefore acceptable in a 2D spectrum. Information content in a 2D arises from crosspeaks, which are formed through evolution (during t 1 ; in COS, 1 H- 1 H J-coupling is the useful component that evolves) and mixing of magnetization (here done by the 2 nd 90 pulse). Fundamental questions in 2D NMR then become how do crosspeaks arise, how intense are they (what needs to be optimized to insure that they are sufficiently intense), and what else needs to be optimized (including resolution) to minimize experiment times. I. Resolution in absolute-value COS: (this is the most common type of COS run in our lab) [Any of the four samples mentioned above will work fine for section.] In this HW, we ll explore the issues of resolution and crosspeak intensity in COS 2D NMR. Resolution is (relatively) easy to understand: it is controlled by the acquisition time in the 2 nd dimension, just as it is in 1D NMR. In the detected dimension (t 2 F2): AQ = (number of points acquired) (time per point) = TD DW [1] The tricky part is Nyquist s theorem: SW = 1/(2DW), so AQ = TD / (2 SW). In the indirect dimension (t 1 F1), we just label everything with a one: AQ1 = TD1 / (2 SW1) [2] Annoyingly, you can t type the parameters this way into TopSpin, but they show up in columns (1 after 2, of course!) in the ACQUPARS (eda) panel. For a 2D COS spectrum: digital resolution ~ 1/AQ1 [3] Note that in normal COS data, SW = SW1 and TD1 < TD. TD1 SW1 AQ1 The digital resolution in COS spectra is dominated by the indirect dimension parameter TD1. I.e., the number of 1 H spectra acquired to form the 2D stack is a crucial parameter during COS experiment setups. [The other three crucial parameters are COS-type selected, D1, and NS.] We re now ready to play around with resolution. In this section, we ll use the simplest, most common form of COS utilizing the experiment (parameter set) COSGPSW.

3 HW#9: 2D NMR, COS Pg. 3 One might wonder how small TD1 can be: experiments get shorter in total time linearly with TD1. a) Acquire a very low resolution cosy: try an experiment with 0.5 to 1 ppm resolution (comute TD1 to match this resolution from eqns 2 and 3). Assume your sweep width will be 8 ppm. D1 Does this make any sense? One might well think not are you kidding me; 1 ppm resolution?? and in general that negativity would be correct. ou ll of course work the data up and see what you think based on spectra. But note how fast the experiment can be done: you re asking for just a few 1 H spectra to be acquired. I (your puny lab director) definitely prefer NS=2, but NS=1 works ok, and is the correct choice when rapid acquisitions are required. Perhaps monitoring of a moderately fast reaction can be done with COS spectra, providing useful information? Maybe There are better methods for taking fast data, and we ll discuss some of those briefly in lecture. Using those other methods, I have seen useful applications of fast TOCS (a similar experiment to COS), as well as HSQC. b) Now acquire a standard cosy (again, using COSGPSW). Calculate the digital resolution for this spectrum. D2 If you find it amazing even this improved resolution works, you re not alone in your first impressions. For years after COS was first invented, prominent NMR spectroscopists believed that TD1 had to be larger than the value used here (128). It took much empirical evidence, and more precise understanding of the COS mixing and evolution to convince otherwise. We ll investigate these issues further in the next section. The key requirement for digital resolution is that it resolve different multiplets. The ability to resolve coupling within the multiplet is removed by the experiment itself: crosspeaks tell us about the coupling, not measurement of the coupling itself. Even so, it is important to keep in mind that TD1=128 is not sufficient for many research situations, so TD1 is a critical parameter to optimize. c) Last in this section is to acquire a high-resolution spectrum, by improving the digital resolution by a factor of four over that for the standard setup. Running this spectrum with NS = 1 is acceptable (although for a research compound, you should leave NS 2; it gives much better data quality). In research involving complex compounds, TD1=512 would be fairly common. Going to TD1=1024 would not be unusual. TD1=1600 and perhaps occasionally 2048 are also used when multiplets are strongly overlapped. [NOTE: Going to such large values with HSQC is not recommended, as will be discussed in the next lab.] 123 Work up the 3 COS spectra acquired above. Spend some time with the processing, using the following steps as a guide: a) In properties (double-left-click the spectrum background), I prefer 40 contours (positive and negative), and Scaling=1.2 as defaults for 2D spectra. This will slow down some operations, so you may need to back off to fewer contours on occasion if things get too slow. b) The Fourier transform sizes in both dimensions need to be identical in order to symmetrize the data (see step f below). This can be checked using Zero Filling and LP within the button. c) The apodization (W) should be set to Sine Square at 0 in both dimensions; click to switch between the two dimensions. For F1, the First Point = 0.5 is normal.

4 HW#9: 2D NMR, COS Pg. 4 d) Set traces (as done in an earlier lab) by loading the proton spectrum with the COS into the same document. Then click on and Setup. e) The COS should look fine at this point. But occasionally MNova does not put the data into magnitude mode. Expansion on a section of the spectrum will then show many positive-tonegative transitions, as shown below as Phased. If so, click and choose Magnitude along F2. our display will then look as shown in the middle (as long as you re displaying a Contour Plot (in ). f) Often the data is improved by symmetrizing it (be wary of symmetrization artifacts; compare spectra prior and after processing). Perform by implementing thru the menu selections: Processing Symmetrize Cosy-Like. Phased Magnitude Mode Symmetrized Plot each spectrum as.pdf and.mnova and upload. II. Crosspeak intensities Long-Range COS: So what s the deal with the crosspeaks? ou should have observed that most crosspeaks are least intense in the poorly resolved (first) spectrum you acquired above. But they won t consistently get larger when going from TD1=128 to TD1=512; some will, but some will not. Crosspeaks arise from transfer of magnetization from one spin in the first time period to a coupled spin in the 2 nd A B time period, e.g.: I t1 I t2. The transfer is sinusoidal in character: during t 1, the magnetization evolves as: sin I I cos Jt I I Jt [4] A t1 A A B 1 Z 1 During the acquisition (and ignoring chemical shift), this magnetization is detected as: I cos Jt cos Jt I sin Jt sin Jt t2 A B The cos 2 A term (starting as I ) forms the diagonal in the COS spectrum. The sin 2 term (starting as A B I IZ forms the crosspeaks. The larger the sin 2 term gets, the larger the crosspeak at (t 1,F1), (t 2,F2) will be. Now we can see that as t 1 1/2J, crosspeaks intensity grows since sin Jt 1 1. Once past 1/2J, the crosspeaks will get smaller. At t 1 = 1/J, the crosspeak will vanish: sin Jt 1 = 0. The evolution time t 1 is therefore critical to crosspeak intensity. Remember that t 1 is the incrementally increasing separation between the two 90 pulses in a COS. It takes another

5 HW#9: 2D NMR, COS Pg. 5 spectrum to increment t 1 another value of 1/SW1. Getting to large t 1 values costs in increased experiment time. What we want is just enough sin Jt 1 to get the crosspeak well out of the baseline noise of the 2D spectrum, but not by too much. It would take a lot of experimentation, and more detailed analysis to go further with this (already overly wordy) discussion, but hopefully you get the idea. For a range of normal proton-proton couplings, 3-15 Hz, TD1 = 128 to 512 works nicely. For smaller TD1 values (and therefore faster experiments), crosspeaks involving smaller J-couplings may be difficult to unambiguously identify. In such cases, increasing NS would help, but increasing TD1 another option. Larger NS gives somewhat better overall signal-to-noise; larger TD1 definitely improves resolution. D3 What about couplings smaller than 3 Hz? Suppose you re interested in verifying a W coupling of 0.8 Hz. From the analysis above, what would t 1 need to equal? If SW=SW1=8ppm, what would TD1 have to equal? [Note: Since we re applying sinebell apodization, the 1/2J point needs to be in the middle of the acquisition, not at the end. That means you need AQ=1/J for maximum crosspeak intensity. ouch ] For a 2 scan COS, how long would the experiment take? (ou could use the spectrometer to estimate this, or realize that D1 NS TD1 is a good estimate.) The key to all this is the rather extreme value you should have arrived at for TD1 (see eqn 2 above). [TD1=12,000 on a 600 when the sinebell apodization correction is included!] One way around this issue is simple: acquire spectra when t 1 is closer to 1/2J. This is the essence of the long-range COS experiment. If we put in a delay D4, typically ranging 50 to 200ms (for small molecules, up to 500ms works fine), crosspeaks arising from small J-couplings will get larger. Long-range COS sequence 4 Acquire a long-range COS with D4=0.2s, TD1=128, and NS 2. Process, symmetrize, and plot. Identify a couple crosspeaks that increased in size from the standard COS. III. Phase-sensitive, double-quantum filtered COS: D4 Let s take this in two bites, phase sensitive first. To prove that phase sensitive is better, go to your 1 H spectrum in MNova, and put the spectrum into magnitude mode: each complex point a + ib in 2 2 the spectrum is replaced by a b. ou ll see that that is not good for resolution. So why do magnitude processing at all? Well, we might be lazy and not want to have to phase the spectra. But the result seems to be awful. We can improve it by applying a sinebell apodization. Try it on the 1D 1 H spectrum: using W, switch the apodization to sinebell at 0. ou ll see that this greatly improves the spectrum: the sinebell narrows the long peak tails arising from the magnitude processing. Unhappily, the ability to quantify the data is also seriously degraded. Such processing is therefore not done, and the (relatively simple) process of phasing 1D spectra is done to achieve good resolution and quantitative spectra. D5 Why then do COS as absolute value (magnitude mode)? Turns out it s not so simple to get a phased 2D spectrum of this type. Claridge provides more detail on this if you are interested. But

6 HW#9: 2D NMR, COS Pg. 6 eqn 4 above tells the tale: the diagonal (arising from the I A term) has cos 2 phase, whereas the crosspeaks (arising the I A B I Z term) are sin 2. The two sets of peaks in a COS spectrum will always be 90 out-of-phase: no way to phase a this simple form of COS spectra. The sinebell apodization recovers much of the lost resolution, and magnitude mode processing removes the issue of having to phase. Everyone say ah! There is a more complex version of COS that gets diagonals and crosspeaks in-phase by implementing a double-quantum filter (cool spin-physics! ). This improves resolution, and adds two additionally useful features: DQ-COS: 1) The coupling actively involved in forming a crosspeak can be identified as being anti-phase (plus-minus). Other couplings involved in the two multiplets are inactive, and are in-phase. See discussion and examples in Claridge Figs 5.45 thru ) The double-quantum filter removes large singlets. To create double-quantum magnetization, two coupled spins are required. tert-butyl groups, a large HOD peak, etc., will be removed in this experiment. 5 Acquire a DQ-COS spectrum with TD1=512, and minimum phase cycle (remember to consult the end of the pulse sequence listing). Process, phase (should be close to correct initially), and plot. Zoom in and identify active versus inactive couplings for at least one coupled pair. Upload 5 plots as.mnova and.pdf files, and come to lecture ready to discuss questions raised here.